Design and Implementation of GPU-Based Frequency Domain Analysis Function in the Digital Oscilloscope

With the increasing complexity of digital signals, the digital oscilloscope which only has the time domain analysis function has been unable to meet the user's needs. In the traditional digital storage oscilloscopes, the frequency domain analysis module is based on the FFT algorithm, and it runs on the CPU in a serial manner. In this paper, we propose a frequency domain analysis framework based on OpenGL technology, which significantly increases the analysis speed of signals while reducing CPU occupancy and memory consumption. the CPU is responsible for the overall scheduling and data flow control, and each sub-module of frequency analysis function runs in the GPU in parallel, and data transfer between CPU and GPU is minimized. The experimental results show that the time consumption and memory consumption of the new architecture is more advantageous when using data volumes over 16k for analysis, and the optimization improves significantly as the data volume increases; when using 128M data volumes for analysis, the total time consumption is only 1/10 of that under the original CPU framework, and the time actually spent on computation is only 1/33 of the original time consumption.